A novel forgery detection technique for document images

Digital documents play an essential role in the day-to-day life of every once life. The documents are shared and exchanged in routine work. At the same time, protecting the document from fraud is a challenging task. The proposed method addresses document forgery detection. In this technique, first, we perform a document enhancement process by removing the noise like salt and pepper and Gaussian using various filtering techniques. Then the enhancement of the contrast by using Histogram Equalization and Contrast stretching. Finally, the deblurring is used to recover the document from degraded observations. In the second process, feature extraction is performed by extracting the color, grey, and entropy features, then the quality measures like Mean Square Error (MSE), Peak Signal-to-Noise Ratio (PSNR), and Structure Similarity Index Matrix (SSIM) are also used as features. In addition, we use Optical Character Recognition (OCR) for fetching textual information. The proposed method performs both textual and image authentication. The proposed method achieved good performance for document forgery detection compared to previous techniques in terms of accuracy.